- 學位論文
以語意標籤為基礎之分散式桌面搜尋
Distributed Desktop Search base on Semantic Tag
摘要
現在一般家庭的電腦愈來愈多，儲存的檔案也隨之增加，管理檔案已經是一件麻煩的事情。傳統的桌面搜尋軟體只針對單一電腦搜尋，很少針對多台電腦，如何有效管理多台電腦，將會是未來熱門的需求。而目前主流的桌面搜尋軟體，搜尋結果跟網頁搜尋比起來，效果並不好，主要的原因在於電腦上的檔案並不像網頁與網頁之間有連結，傳統的Page Rank方法並不適用於桌面搜尋，而且網頁搜尋引擎可以根據廣大使用者的Feedback重新排序搜尋結果。 本篇論文的方向著重在如何有效的搜尋多媒體檔案，以及同時搜尋多台電腦的檔案。大略的作法是，一開始將電腦檔案裡面相關的Metadata抽取出來並轉成Tag，如果多個檔案的Metadata內容相同，就關聯到同一個Tag。之後在搜尋時，將Query送到每一台電腦，每台電腦再搜尋相關的Tag，再從Tag搜尋檔案。最後系統收集到各台電腦的搜尋結果，再對結果進行最後的正規化、重新排序搜尋結果。本論文提出的方法對於多媒體檔案的搜尋有顯著的改善，而在分散式桌面搜尋也有初步成果，對於後續的研究有一定的參考價值
並列摘要
Many families have more and more computers now, storage of files have increased. The management of the files has become a trouble. The traditional desktop search software search only for a single computer and rarely for more than one computer. How to effectively manage more than one computer will be the next popular demand. The mainstream of desktop search software’s search results is not good than the web search results. The main reason is there are no links between computer files. Traditional method of PageRank can’t be directly applied to desktop search, and web search engines have feedback based on vast numbers of users to re-sort search results This paper focuses on how to effectively search for files across multiple computers, and how to merge and re-rank search results from more than one computer. First, extract Metadata from computer files and convert into Tag. If many files have the same Metadata, then link to the same Tag. When searching, the system will send query to each computer, and each computer use Semantic search. Finally, the system collect result from each computer, and product final results with normalizing and re-sorting